Rethinking GNNs

Over-squashing, Bottlenecks, and Graph Ricci curvature

Over-squashing is a common plight of Graph Neural Networks occurring when message passing fails to propagate information efficiently on the graph. In this post, we discuss how this phenomenon can be understood and remedied through the concept of Ricci curvature, borrowed from the field of differential geometry.

Michael Bronstein
Towards Data Science
9 min readNov 30, 2021

--

--

--

DeepMind Professor of AI @Oxford. Serial startupper. ML for graphs, biochemistry, drug design, and animal communication.